huawei-csl/Apertus-8B-2509-4bit-SINQ
Text Generation
•
5B
•
Updated
•
7
None defined yet.
SINQ: Sinkhorn-Normalized Quantization for Calibration-Free Low-Precision LLM Weights
Top-Theta Attention: Sparsifying Transformers by Compensated Thresholding